- Joined
- Oct 28, 2017
Last edited by a moderator:
-
Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!
Just when you think the OC is stable...
- Thread starter mackerel
- Start date
That is good. We were talking about the more recent processors from LGA 1151.
OP
- Joined
- Mar 7, 2008
- Thread Starter
- #24
To give a system status update... no new errors in the last 3 days. To recap, I did two changes at the time, move the GPU to a different slot to give more space around the CPU cooler, and also change the software from running 4x single thread tasks to 2x two thread tasks. Like all good troubleshooting, I didn't only change one variable so I don't know which one did it. As the thermal images didn't reveal any obvious hot spots I'm still inclined to think it is an edge case scenario with the CPU that is sidestepped by running the work differently. Although runtimes are reduced by running two threads per task, it isn't perfect scaling so I lose around 12% throughput by doing this. I had to do it for other reasons anyway. The project has Top Gun style "no points for 2nd place" reporting of prime numbers, so there is an advantage to shorter turnaround times. More and more people are enabling it so I have to keep up.
Could the chip possibly be degrading over time?
"There are known transistor degradation mechanisms such as gate-oxide
breakdown and hot-electron effects that slowly change transistor
performance. This can slowly degrade timings for signals across chip
and eventually cause the chip to not work. Or it can result in a bit
flip, if it happens in the chip's cache"
"There are known transistor degradation mechanisms such as gate-oxide
breakdown and hot-electron effects that slowly change transistor
performance. This can slowly degrade timings for signals across chip
and eventually cause the chip to not work. Or it can result in a bit
flip, if it happens in the chip's cache"
have you changed your bios version since the original test? same voltage? it may just need more volts, vcore or vccio/sa or dram.
I had this happen years ago with a socket 478 p4, I had to keep raising the voltage every 6 months or so, may or may not be the same in your case. What speed and what voltage do you run it at?
I had this happen years ago with a socket 478 p4, I had to keep raising the voltage every 6 months or so, may or may not be the same in your case. What speed and what voltage do you run it at?
Last edited:
OP
- Joined
- Mar 7, 2008
- Thread Starter
- #30
Timings are NOT stock. They're about 95% optimised as far as it'll go, with stability testing throughout. To repeat, the tasks done should be small enough not to hit system ram much. Only weeks ago, with no changes to configuration, I was running bigger tasks that hit the ram hard without error. Since I changed the system to work on 2x 2 thread tasks (instead of 4x 1 thread) it hasn't resulted in any errors so I'm leaving it.
